The most serious evidence-based critique of India’s reservation system is not that it fails to help disadvantaged groups. The data shows that it has helped. The sharper problem is one of distribution. Within each group that qualifies for reservation, the benefits tend to concentrate among the households that are already better off, better informed, and better positioned to compete. The families with the least, the ones the policy was designed to reach first, are often the last to arrive, if they arrive at all.
This is a reform argument, not an argument for abolition. A policy can be justified in principle and still be poorly targeted in practice. Understanding where reservation benefits actually land, and why, is the first step toward making them reach the people who need them most. This is Part 2 of a five-part series examining reservation on the evidence. Part 1 looked at what the data shows reservation has and has not achieved. This part examines the internal distribution problem that sits at the center of almost every honest reform proposal.
What the creamy layer actually means
The phrase “creamy layer” entered Indian constitutional law through the Supreme Court’s nine-judge decision in Indra Sawhney v. Union of India in 1992. While upholding 27 percent reservation for Other Backward Classes, the Court held that the socially advanced and wealthier members of those classes, the so-called creamy layer, should be excluded from reservation benefits. The reasoning was straightforward. If the goal of reservation is to lift those held back by social and educational backwardness, then people within a backward class who have already escaped that condition should step aside so that the benefit flows to those still held back.
The Court left the government to define the boundary. Over time this became an income and status ceiling. Children of parents above the ceiling are treated as creamy layer and are not eligible for OBC reservation. The ceiling has been revised upward several times and currently sits at an annual family income of about 8 lakh rupees, alongside exclusions tied to high constitutional office and senior government rank. Certain kinds of income, notably salary and agricultural income, are treated differently from business and professional income when the ceiling is applied, which itself has been a source of dispute.
One point is central and often misunderstood. The creamy layer test in Indra Sawhney was applied to OBCs. It was not applied to Scheduled Castes and Scheduled Tribes. The Court treated SC and ST status as arising from a distinct history of untouchability and exclusion that income alone does not erase. That distinction held for more than three decades and has only recently been reopened, as discussed below.
How the ceiling has moved, and where it leaks
The creamy layer boundary is not a fixed number. It has been raised repeatedly to keep pace with rising incomes, moving from 1 lakh rupees a year when it was first set in 1993 to 2.5 lakh, then 4.5 lakh, then 6 lakh, and to about 8 lakh rupees a year in 2017. Each revision widens the pool of OBC families that remain eligible, which is popular with beneficiary communities and is resisted by those who want the benefit reserved for the poorest.
The definition also treats income unevenly, and this is where enforcement leaks. Salary and agricultural income are not counted toward the creamy layer ceiling in the same way that business and professional income are. The original logic was that a salaried government post already reflects a certain status, so the test should look at other markers. In practice the carve-out means two families with identical living standards can be treated differently depending on where their income comes from. Successive governments have debated whether to count gross salary toward the ceiling, and the matter remains unsettled. The result is that the creamy layer, even where it applies, does not cleanly separate the advantaged from the disadvantaged.
A further gap is verification. Eligibility rests on income and status certificates issued at the local level, and the system for checking them is thin. Where documentation is weak, the households best able to obtain a favorable certificate are again the better-connected ones. A rule is only as good as the administrative machinery that applies it, and here the machinery is uneven across states.
The concentration of benefits: the strongest data point
If there is a single finding that reform advocates return to, it is the degree to which OBC reservation benefits are captured by a small number of dominant communities within the OBC category.
The Rohini Commission, constituted in 2017 to examine the sub-categorization of OBCs, studied the distribution of central OBC reservation across communities. Its analysis, as widely reported, found that a large majority of reserved benefits were flowing to a small minority of OBC castes. According to figures attributed to the Commission’s work, roughly 97 percent of central OBC reserved jobs and educational seats were going to about 25 percent of OBC castes and sub-castes. Nearly 1,000 of the roughly 2,600 OBC communities on the central list had close to zero representation in reserved posts and seats.
The meaning of these numbers deserves care. They do not show that reservation failed. They show that a category built from thousands of distinct communities is not internally uniform. Some communities within the OBC list entered the modern economy earlier, had more land, more literacy, and more social capital, and were therefore better placed to use reserved seats the moment they opened. Other communities on the same list started further behind and have barely accessed the benefit at all. A single quota, shared across a group this varied, will naturally be drawn down first by those already closest to the opportunity.
| Feature | OBC reservation | SC / ST reservation (historically) |
|---|---|---|
| Creamy layer excluded | Yes, since 1992 | No, until the 2024 reopening |
| Basis of backwardness | Social and educational | History of untouchability and exclusion |
| Income ceiling applied | Yes, about Rs 8 lakh per year | Not applied historically |
| Documented internal concentration | High, per Rohini Commission findings | Debated, prompting the sub-classification question |
The table simplifies a contested area, but it captures the asymmetry that recent litigation has tried to address.
A pattern seen beyond India
Internal concentration is not unique to India, which suggests it is a structural feature of group-based preference rather than a flaw specific to one country. Malaysia offers a clear parallel. Its New Economic Policy, launched in 1971, extended broad preferences to the Bumiputera majority in education, employment, and ownership. Studies of the policy over five decades credit it with a large fall in Bumiputera poverty and the growth of a sizeable middle class. The same studies also record a persistent rise in inequality within the Bumiputera group, and the emergence of a small, politically connected elite that captured a disproportionate share of the gains.
The lesson that travels is this. When a benefit is defined by membership in a large group, the members who are already ahead tend to absorb it first, and the gap inside the group can widen even as the group’s average improves. Recognizing this pattern is what motivates targeting reforms everywhere the approach is used, from Kuala Lumpur to New Delhi.
Sub-classification and the 2024 turning point
The question of internal concentration is not limited to OBCs. States have long argued that Scheduled Castes are also not a single homogeneous block, and that a few communities within the SC list capture most of the reserved benefit while others remain excluded. The legal problem was whether a state could divide the SC list into sub-groups and give the most disadvantaged among them a protected share.
On 1 August 2024, a seven-judge Constitution Bench of the Supreme Court answered that question. In State of Punjab v. Davinder Singh, by a majority of six to one, the Court held that states do have the constitutional power to sub-classify Scheduled Castes for the purpose of reservation. The majority reasoned that Scheduled Castes are not homogeneous, that some communities within the list are more disadvantaged than others, and that a state may identify the most backward among them and ensure they receive a fair share. The decision overruled the Court’s earlier 2004 ruling in E.V. Chinnaiah v. State of Andhra Pradesh, which had treated the SC list as a single indivisible class. Coverage of the judgment is available through LiveLaw and other legal outlets.
The judgment went further and touched the creamy layer itself. Several judges observed that states should also evolve a policy to identify a creamy layer within Scheduled Castes and Scheduled Tribes, so that the most advanced members do not crowd out the rest. The observation was not framed identically to the OBC rule. The suggestion was that within the SC quota, the more advanced would take a lesser and relative share rather than being removed from the quota entirely.
This part of the ruling is genuinely contested, and an honest treatment presents both sides.
- The case for it. Supporters argue that if the goal is to reach the worst-off, then a family that has already secured stable government employment and higher education across a generation is not the same as a landless first-generation household, even within the same caste. Directing the benefit toward the latter is consistent with the original purpose.
- The case against it. Critics, including civil-liberties groups such as the Citizens for Justice and Peace, argue that the creamy layer logic imported from OBCs does not fit SC and ST realities, because untouchability and social stigma are not dissolved by a rise in income. A Dalit professional can earn well and still face exclusion that an equally paid person from a dominant caste never encounters. On this view, applying an income test to SC reservation misreads what the reservation is compensating for.
Both positions are serious. The disagreement is not about whether the poorest should be reached, but about whether an income filter is the right instrument for a disadvantage rooted in social status rather than money.
Intergenerational capture
Underneath the legal debate sits a simpler worry that shows up across all reserved categories. Once a family uses reservation to secure education and a stable job, its children start life with advantages that most families in the same category still lack. Those children are then well placed to use reservation again. Over two or three generations, a set of families can convert an initial benefit into a durable head start, while first-generation households from the same community never cross the threshold to compete for the seat at all.
This is not a moral failing of the families involved. Anyone would use a legal benefit their children qualify for. It is a design feature of a system that measures eligibility by group membership rather than by whether the specific household has already been lifted. The creamy layer rule for OBCs is one attempt to interrupt this cycle. Whether it should be extended, and in what form, to other categories is the live question.
There is a counterweight worth stating fairly. For SC and ST families in particular, a single generation of advancement does not always translate into security. Social stigma can follow a family that has risen economically, and a downturn, a health shock, or discrimination in hiring can pull a first-generation professional’s children back. This is precisely why critics resist a simple income filter for SC and ST reservation. The honest position is that intergenerational capture is real and worth correcting, and that the correction has to be sensitive to the difference between economic mobility and social acceptance, which do not always move together.
Reform options and their trade-offs
The distribution problem has several possible responses, and each involves trade-offs. None of them requires ending reservation.
- Enforce the creamy layer more strictly for OBCs. The ceiling exists, but enforcement is uneven, and the treatment of different income types creates loopholes. Tightening the definition and closing gaps would push benefits toward poorer OBC households. The trade-off is administrative complexity and the risk of excluding genuinely backward families through crude income lines.
- Sub-categorize within the group. Dividing a large category and reserving a protected share for the most backward sub-communities directly addresses concentration. Tamil Nadu’s Most Backward Classes category and the newly permitted SC sub-classification both follow this logic. The trade-off is the political difficulty of deciding which communities go into which tier, and the data burden of doing it fairly.
- Improve the data. Rational targeting requires knowing the current condition of each community, which is why the caste census question matters here. Without recent, disaggregated data, both quotas and creamy layer lines rest on estimates that are decades old. The trade-off is that a census is politically charged and slow, and its findings can unsettle existing arrangements.
- Add a needs test inside the group. Some proposals keep caste as the gate but rank applicants within the reserved pool by household deprivation, so the poorest members rise first. This is examined in detail in Part 3 of this series, on caste versus economic criteria.
Each of these is a way to keep the policy and fix its aim. The choice among them is a question of evidence and administration, not of whether disadvantaged groups deserve support.
What states have already tried
Several states have not waited for a national settlement and have built their own answers to the concentration problem. Their experience is the closest thing to a live experiment.
Tamil Nadu splits its Backward Classes quota and carves out a separate Most Backward Classes and Denotified Communities share, so that the communities furthest behind are not forced to compete head to head with the more advanced Backward Classes for the same seats. The state has run this tiered design for decades, and it is one reason its reservation architecture is studied so closely.
Andhra Pradesh attempted to sub-divide its Scheduled Caste quota in the early 2000s, which produced the litigation that ended in the 2004 E.V. Chinnaiah ruling against sub-classification. Punjab gave first preference within the SC quota to the Valmiki and Mazhabi Sikh communities, a practice that was challenged and eventually reached the seven-judge bench in 2024. In other words, the question the Supreme Court settled last year had been building from state-level attempts to fix concentration for close to twenty years.
The pattern across these cases is consistent. States closest to the ground could see that a single quota was not reaching the most excluded communities, and they tried to redirect it. The legal system spent two decades deciding whether they were allowed to. With that question now answered, the harder administrative work of drawing fair sub-categories begins.
Why this matters for the institutions
A reservation system that reaches the same advantaged families repeatedly produces two problems at once. It leaves the poorest members of a beneficiary group behind, which defeats the policy’s own purpose. And it hands critics an easy argument, because visible benefit capture by better-off households is used to attack the entire framework. Targeting the benefit better is therefore not only fairer to the intended recipients. It is also the strongest defense of the system against the claim that it has drifted from its purpose.
The goal, stated plainly, is that a reserved seat should reach a student or worker who could not have reached it otherwise. When the seat instead goes to someone whose family had already made it, the policy has spent a scarce opportunity without moving anyone forward. Fixing that is the work of reform.
A final point deserves emphasis, because it is often lost in the noise. Correcting benefit capture is not the same as shrinking the commitment to social justice. A policy that reaches the poorest within each group does more real work per seat than one that recycles advantage, so better targeting raises the return on every reserved place rather than lowering it. The reform argument in this article is therefore an argument for making reservation work as intended, not an argument for winding it down. The distinction matters, because the same evidence of capture is read by some as a reason to expand support to those still left out, and by others as a reason to withdraw it. Which lesson a society draws is a political choice, not a statistical one, and it is worth making that choice with the data in full view.
Frequently asked questions
Does the creamy layer rule apply to Scheduled Castes? Historically no. The 1992 Indra Sawhney judgment applied the creamy layer test to OBCs but exempted SCs and STs. In August 2024, the Supreme Court in State of Punjab v. Davinder Singh suggested that states should identify a creamy layer within SCs and STs as well, though in a form different from the OBC rule. This is now a live policy question rather than settled practice.
What is the current creamy layer income ceiling for OBCs? The ceiling is an annual family income of about 8 lakh rupees, revised upward over time, alongside exclusions based on high public office and senior government rank. Different income types are treated differently when the ceiling is applied, which has produced ongoing disputes.
What did the Rohini Commission find? Reporting on the Commission’s analysis indicated that central OBC reservation benefits were heavily concentrated, with roughly 97 percent of reserved jobs and seats going to about a quarter of OBC communities, and close to 1,000 communities on the list receiving almost nothing. The figures illustrate how unevenly a single quota is drawn down across a very varied category.
Does concentration of benefits mean reservation has failed? No. It means a broad category is internally unequal, so the benefit reaches the better-placed members first. That is a targeting problem to be corrected, not evidence that the policy should be removed.
Is sub-classification the same as abolishing reservation? No. Sub-classification keeps the overall quota and divides part of it to protect the most backward communities within a group. It is a redistribution inside the reserved category, intended to reach those currently left out.
Do other countries face the same concentration problem? Yes. Malaysia’s decades-long preference policy for its Bumiputera majority reduced group poverty while inequality within the group rose and a connected elite captured an outsized share. The pattern suggests that benefit concentration is a structural feature of group-based preference, which is why targeting reforms appear wherever the approach is used.
What is the single most useful reform? There is no single answer, but better and more recent data underpins every other option. Reliable, disaggregated information on the current condition of each community is what allows creamy layer lines, sub-quotas, and needs tests to be set fairly rather than by guesswork. This is why the debate over counting caste in the census is not a side issue but the foundation for every targeting decision that follows.